Towards Decoding Selective Attention from Single-Trial EEG Data in Cochlear Implant Users

Background: An attended speech source in normal hearing (NH) listeners can be decoded by single-trial EEG data.

Objectives: This method might have the potential to be used to steer noise-reduction and sound coding algorithms in cochlear-implant (CI) users improving the signal-to-noise ratio for a given attended target speaker. However, the decreased spectral resolution and the electrical artifacts introduced by a CI may impair decoding accuracy. The study aims at investigating whether selective attention can be decoded in CI users.

Methods: To investigate the effect of spectral smearing on the decoding accuracy, we tested 12 NH listeners using vocoded sounds. To evaluate the additional effect of the CI artifact, we moreover tested 12 CI users. For the task, speech from two audio books was presented through insert earphones to NH listeners and via direct audio cable to CI users. Participants were instructed to attend to one out of the two concurrent speech streams while a 96-channel EEG was recorded. Reconstruction performance was evaluated by means of parametric correlations between the reconstructed speech and both, the envelope of the attended and the unattended speech stream.

Results: Results show the feasibility to decode selective attention by means of single-trial EEG data not only in NHs with a vocoder simulation, but also in CI users if enough training data are available. It seems that limitations in decoding selective attention in CI users are more influenced by the lack of spectral resolution than by the artifact caused by CI stimulation.